Validation of Environmental Exposure Visits Using an Emergency Department-Based Electronic Syndromic Surveillance System

BACKGROUND:  Emergency department-based electronic syndromic surveillance systems (EDSSS) are used to detect and monitor public health problems.  Emergency department (ED) chief complaint data may be used to capture potential non-infectious as well as infectious health-related visits.  The Pennsylvania Department of Health (PADOH) recently developed chief complaint-based syndrome definitions to detect visits associated with weather events and other environmental exposures.  Definitions were validated using data from the subset of EDs that also submit discharge diagnosis data.

METHODS:  EDSSS visits occurring from June 30, 2013 to July 4, 2015 (105 weeks) were included in the study.  Of these, 1,062,808 visits (12.4% of all ED visits during that time period) contained valid Ninth Revision of the International Classification of Diseases (ICD-9) codes in the discharge diagnosis field.  Syndrome definitions in use by other health departments were reviewed and customized definitions were developed for carbon monoxide exposure, extreme cold injury, extreme heat injury, acute asthma, heroin exposure, and tick bite.  ICD-9 codes corresponding to each syndrome were also defined. Within the subset of visits that had ICD-9 diagnosis codes, both chief complaint and ICD-9 definitions were applied, and predictive positive value (PPV) and sensitivity were calculated for each syndrome.  To evaluate the system-wide representativeness and validity of the chief complaint definitions, they were also applied to all ED visits.  The numbers of weekly visits detected by each chief complaint syndrome definition were compared to corresponding counts obtained from the ICD-9 definitions using the Spearman rank correlation coefficient (r_s).

RESULTS:  Sensitivity values (the percent of visits identified by ICD-9 as belonging to a particular syndrome that were captured by the chief complaint definition) were <40% for all syndromes except for carbon monoxide exposure (67.4%). However, PPV was 70% or greater for all syndromes except tick bite (61.0%).  All Spearman rank correlation coefficients derived from comparisons of chief complaint-derived syndromes to ICD-9 results were statistically significant (p<0.001) and ranged from moderate to high for carbon monoxide exposure (r_s = 0.61), acute asthma (r_s = 0.62), heroin exposure (r_s = 0.62), extreme heat injury (r_s = 0.71), extreme cold injury (r_s = 0.74), and tick bite (r_s = 0.74). 

CONCLUSIONS:  Our findings support the validity of using EDSSS to detect a variety of environmental exposures.  Application of these validated methods will enhance the ability of PADOH staff to identify surveillance pattern changes related to environmental exposures in a timelier manner, as well as facilitate situational awareness, response and public messaging.